Development of screening methods for transketolase activity and substrate scope

Article abstract of DOI:10.1016/j.tetlet.2011.12.010

A new GC-based method is developed for the screening of activity and stereoselectivity of a wide range of polyols as substrates of transketolase (TK). Compared to previous screening methods, it shows higher reproducibility, sensitivity and range of detection. In combination with HPLC screening, it can be used efficiently to test mutant libraries obtained by directed evolution methods.

Full text of DOI:10.1016/j.tetlet.2011.12.010

Tetrahedron Letters 53 (2012) 790–793
Contents lists available at SciVerse ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Development of screening methods for transketolase activity
and substrate scope
⇑
Adeline Ranoux, Isabel W.C.E. Arends, Ulf Hanefeld
Technische Universiteit Delft, Gebouw voor Scheikunde, Afdeling Biotechnologie, Julianalaan 136, 2628 BL Delft, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
A new GC-based method is developed for the screening of activity and stereoselectivity of a wide range of
polyols as substrates of transketolase (TK). Compared to previous screening methods, it shows higher
reproducibility, sensitivity and range of detection. In combination with HPLC screening, it can be used
efficiently to test mutant libraries obtained by directed evolution methods.
Received 12 August 2011
Revised 21 November 2011
Accepted 2 December 2011
Available online 8 December 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Transketolase
Screening methods
Substrate scope
Selectivity
Mutant libraries
Transketolase (TK) catalyses the asymmetric formation of C–C
bonds. This is of great interest for chemical synthesis, especially
for the preparation of biologically active compounds. The natural
substrates of this enzyme are phosphorylated polyols (Scheme 1),
but various examples of non-phosphorylated short chain sub-
strates have been described.1
Therefore, we proposed a medium throughput screening method
for the coupling reactions between a long chain hydroxylated alde-
hyde acceptor (any sugar, D-ribose or D-glucose for example) and a
keto acid donor (Scheme 2) based on GC analysis after derivatisation.
Indeed, this method has four major advantages: accuracy and sensi-
–3
15
tivity, the possibility of studying the stereoselectivity of the enzy-
Directed evolution and related mutation methodologies4–6 can
be used as tools to expand the scope of TK, notably towards new
non-phosphorylated substrates.7 They require generally high
throughput screenings of very large libraries of mutants. However,
the recent development of more rational approaches such as Itera-
matic reaction and a wide range of substrates can be utilised.
Moreover, the required derivatisation of the substrates in order to
obtain volatile compounds for the GC method suppresses the forma-
tion of different isomers (regiomers, anomers, furano or pyrano
forms), and thus highly facilitates the analysis.
8
tive Saturation Mutagenesis enables the size of the libraries to be
An example of a GC-based screening method for TK has already
1
6
limited to 1000–5000 mutants. The development of efficient, sen-
sitive and reproducible screening methods for these libraries is a
successfully been implemented by Smith et al. to study the enanti-
oselectivity of the coupling of propanal and lithium hydroxypyru-
vate. The detection levels were shown to be very low (within 1–
2%). However the derivatisation protocol, leading to the formation
of acetate derivatives, requires extraction of the products, which
cannot be applied in the case of polyhydroxylated compounds.
Therefore, we proposed a different approach adapted to our targeted
substrates. The chosen derivatisation method leads to oxime-TMS
derivatives in a straightforward two-step, one-pot procedure.
Despite its accuracy, this GC-based screening method remains
difficult to apply for very large libraries of mutants. Thus, for more
efficiency, this accurate GC-based screening method was combined
with another faster screening approach allowing for the rapid
selection of the TKs active after mutation. The coupling of hydroxy-
pyruvate (HPA) and glycolaldehyde (GA) (Scheme 3) was used as it
represents a model assay to measure the specific activity of TK
with non-phosphorylated compounds.
9,10
key step to identifying rapidly, successful mutations.
The
screening approach should ideally be as general as possible, appli-
cable to a wide range of substrates (Scheme 2), and provide accu-
rate information on all aspects of the reaction, characterising both
activity and selectivity (including enantioselectivity). The use of
9
6-well plates should ensure application as a medium throughput
assay.
To date, colorimetric or fluorogenic detection methods have
been used for TK screening. However, these methods are limited
12,13
to a restricted range of substrates
and range of detection. Finally, they do not give access to equally
important information such as stereoselectivity.
and also have low sensitivity
1
4
⇑
Corresponding author. Tel.: +31 (0) 15 2789304; fax: +31 (0) 15 2781415.
E-mail address: u.hanefeld@tudelft.nl (U. Hanefeld).
0
040-4039/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2011.12.010

A. Ranoux et al. / Tetrahedron Letters 53 (2012) 790–793
791
H
OH
O
OH
O
HO
HO
H
HO
HO
O
Transketolase
Mg2+, TPP
OH
OH
O
OH
+
OH
+
HO
OPO32-
HO
OPO32-
OPO32-
OPO32-
Scheme 1. Transketolase catalyses the coupling of ribose-5-phosphate and xylulose-5-phosphate in nature [using cofactors, Mg2+ and thiamine pyrophosphate (TPP)].
tures were derivatised before GC analysis by forming TMS-oximes
in the acyclic form. The protocol used is rapid and does not require
R'
CO 2
O
R
H
LiOOC
R
R'
multiple treatment steps as is the case for alditol acetate or aldon-
OH
+
21,22
Transketolase
onitrile acetate derivatisations.
Moreover, it was systemised,
O
O
OH
OH
2
+
automated and adapted to be carried out directly in 96 deep-well
plates of 2 mL.
Mg , TPP
Scheme 2. General scheme for the coupling of non-phosphorylated compounds.11
The crude enzymatic samples were derivatised directly after the
removal of water without any pre-purification. Indeed, it has been
shown that the presence of salts and proteins does not affect the
2
3
derivatisation reaction. Similarly, the high volatility of silylated
products and by-products makes any post-derivatisation purifica-
OH
TK, Mg2+, TPP
O
LiOOC
HO
2
4
+
OH
tion steps superfluous. Thus, the derivatised samples can be di-
rectly injected into the GC instrument. The two derivatisation
steps (Scheme 5) were carried out in one-pot within 1 h. The sily-
lating agent, N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA),
and 1% trimethylsilyl chloride (TMSCl) were chosen for their high
reactivity especially with water. Indeed, this property eliminates
moisture that could degrade the derivatised compounds: water
will react with the excess BSTFA rather than the sugar derivative.
HO
OH
Tris-HCl (50 mM)
25 °C
O
O
GA
HPA
Scheme 3. Specific activity measurements by HPLC of a simple coupling of non-
1
7
phosphorylated compounds; formation of erythrulose.
1st screening
2nd screening
25
Addition of TMSCl further promoted the reaction. An internal
standard (IS; glucose) was introduced into the reaction mixture
prior to derivatisation. It underwent the same derivatisation reac-
tion thereby verifying the outcome.
Assay :
Preselected
active mutants
(100 - 300)
Assay :
Library of
Hit mutants
(20)
GA + LiHPA
Ribose + LiHPA
1
000 mutants
Derivatisation
HPLC analysis
GC analysis
The calibration (Fig. 1) for both steps (Schemes 3 and 5) was
carried out for products and reagents (commercial compounds if
available or isolated product). Sedoheptulose was isolated from
the enzymatic reaction (isolated yield: 81%). A good reproducibility
for both screening methods was demonstrated when tested on dif-
ferent wild type enzymes derived from Escherichia coli or yeast (Ta-
ble 1). Moreover, the GC screening method was demonstrated to
detect very low bioconversion (>1%), much lower than previously
described methods (e.g., a colorimetric assay using tetrazolium
Scheme 4. Combined complementary screening methods; preselection of the
active enzymes then defined selection of the most active mutants towards the
targeted substrate coupling.
This coupling has been used in a previously described HPLC-
based screening method.18 Moreover, it has been improved
by the implementation in a microplate assay; application of high
flow rates thus enables the cycle time to be reduced drastically.
Therefore, combined complementary screening methods were
developed (Scheme 4). Based on previously established meth-
19,20
1
2
red was limited to >8% bioconversion ).
Libraries of transketolase mutants obtained by saturation site-
directed mutagenesis were subjected to this dual screening proto-
col. It allowed us to screen the TK activity for couplings of different
non-phosphorylated substrates. It also permitted the verification of
the compatibility and complementarity of both screening methods.
It was shown that the first screening method led to the selection of
10–30% of the mutants, whereas the second defined 2–5% of hits
that all belonged to the first selection. The combination of these
two methods, a first fast activity test, followed by a more defined
screening, appeared to be the most efficient and adaptable protocol
(Scheme 4). Moreover, used in combination with parallel HPLC sys-
1
8,19
ods,
the HPLC method was applied as a rapid pre-screening,
allowing a primary selection of active enzymes towards non-phos-
phorylated compounds. A protocol was implemented at high flow
rate (0.8 mL/min) to reach the final method of 10 min, while allow-
ing direct sampling from 96-well plates.
Next, the preselected enzymes were submitted to a more sensi-
tive second screening with the coupling of longer-chain substrates
such as the reaction of D-ribose with HPA (Scheme 5, other polyols
could potentially be used). For this screening, the reaction mix-
TK Yeast,
Mg2+, TPP
O
O
OH
OH
OH
OH
OH
OH
OTMS OTMS OTMS OTMS
i) HONH2.HCl, Pyr.
HO
+
COOLi
H
ii)
Tris-HCl (50 mM)
Si
O
OH
OH
O
OH
OH
N
OTMS OTMS
25 °C
Si
TMSO
N
HPA
D-Ribose
D-Sedoheptulose
F C
BSTFA
1% TMSCl
3
+
Scheme 5. Second screening assay: coupling of D
-ribose and HPA followed by derivatisation26—GC analysis.27

7
92
A. Ranoux et al. / Tetrahedron Letters 53 (2012) 790–793
Area Ratio(x0.01)
Area Ratio
B
A
4.0
0
0
0
0
.15
.10
.05
.00
Y = aX + b
3.0
-
-5
3
Y = aX + b
a = 7.58.10
a = 7.98.10
b = 7.91.10
r = 0.9952
-
3
-
3
b = -1.35.10
r = 0.9945
2
1
0
.0
.0
.0
2
r = 0.9905
2
r = 0.9890
0.0
5.0
10.0
15.0
Conc. Ratio
0.0
1.0
2.0
3.0
4.0
Conc. Ratio
Figure 1. Calibration curves for the GC screening: FID response (corresponding to the peak area ratios to IS) as a function of the concentration ratios to IS for the substrate, (A)
2
8
D-ribose, and the product, (B) D-sedoheptulose.
Table 1
Results and standard deviations for the GC and HPLC methods applied to wild type enzymes (n.d.: not determined).
Enzyme
Screening HPLC—specific activity
Screening GC—conversion (on 24 h)
Commercial TK yeast
Commercial TK E. coli
Lysate yeast
0.55 (±0.02) U/mg
0.036 (±0.01) U/mg
4.26 (±0.02) U/mL
0.59 (±0.03) U/mL
81% (±4)
n.d.
10% (±2)
n.d.
Lysate E. coli
tems,18,19 it allows the screening of a medium size library (a few
thousands mutants) in 1 week (instead of a few weeks).
300
l
L. The sample can then be either purified by column chromatography,
directly analysed by HPLC for the first screening, or derivatised in view of the
second screening (GC analysis). To prepare the standard solutions, the product
could be isolated as followed: silica gel was added and the reaction mixture
concentrated to dryness under vacuum, then dry loaded onto a flash silica gel
column. Following silica column purification (EtOAc:iPrOH:H O, 7:3:1), the
2
product was isolated generally as a white solid.
In conclusion, automated protocols were implemented to screen
libraries of mutant enzymes. They were tested successfully on com-
mercial TKs and also on in-house wild-type lysates derived from
E. coli and yeast. Significant improvements in terms of reproducibil-
ity, sensitivity and range of detection were achieved. With minor
modifications this assay will also be applicable to other enzymes
1
1
1
1
1
1
2. Smith, M. E. B.; Kaulmann, U.; Ward, J. M.; Hailes, H. C. Bioorg. Med. Chem. 2006,
14, 7062–7065.
3. Sevestre, A.; Hélaine, V.; Guyot, G.; Martin, C.; Hecquet, L. Tetrahedron Lett.
2
003, 44, 827–830.
4. Adams, M. A.; Chen, Z.; Landman, P.; Colmer, T. D. Anal. Biochem. 1999, 266, 77–
4.
1
29
of sugar metabolism, such as aldolases and transaldolases.
8
5. Moldoveanu, S.; David, V. Sample Preparation in Chromatography Vol. 65;
Elsevier: London, 2002.
6. Smith, M. E. B.; Hibbert, E. G.; Jones, A. B.; Dalby, P. A.; Hailes, H. C. Adv. Synth.
Catal. 2008, 350, 2631–2638.
Acknowledgments
The authors gratefully acknowledge financial support from
Shell Global Solutions International B.V. (Dr. Aldo Caiazzo). They
are grateful to the referees for valuable suggestions. The authors
gratefully acknowledge careful proofreading by Professor H. C.
Hailes (UCL, London).
7. HPLC analysis: After 20 min of assay at 25 °C, the reaction is stopped by adding
300 lL of aqueous TFA (trifluoroacetic acid) at 0.2% v/v (1:1 dilution). The
mixture is centrifuged to eliminate the precipitated enzyme, then directly
analysed by HPLC. The analysis was carried out on a HPLC Shimadzu UFLC LC-
20AD equipped with
Shimadzu RI and UV-detectors (210 nm). The mobile phase was an aqueous
solution of TFA at 0.1% with a flow rate of 0.8 mL min . The column was
maintained at 60 °C. The overall method took 10 min. The retention time of the
Ò
a
CarboSep Coregel column (by Transgenomic ),
À1
References and notes
standards was: LiHPA 6.4 min in UV, 6.7 in RI; erythrulose 8.7 min in UV, 8.9 in
RI.
18. Hibbert, E. G.; Senussi, T.; Costelloe, S. J.; Lei, W.; Smith, M. E. B.; Ward, J. M.;
Hailes, H. C.; Dalby, P. A. J. Biotechnol. 2007, 131, 425–432.
19. Miller, O. J.; Hibbert, E. G.; Ingram, C. U.; Lye, G. J.; Dalby, P. A. Biotechnol. Lett.
2007, 29, 1759–1770.
1.
2.
3.
4.
Sukumaran, J.; Hanefeld, U. Chem. Soc. Rev. 2005, 34, 530–542.
Wohlgemuth, R. J. Mol. Catal. B: Enzym. 2009, 61, 23–29.
Turner, N. J. Curr. Opin. Biotechnol. 2000, 11, 527–531.
Hailes, H. C.; Dalby, P. A.; Woodley, J. M. J. Chem. Technol. Biotechnol. 2007, 82,
1
063–1066.
5
6
7
.
.
.
Reetz, M. T. Angew. Chem. Int. Ed. 2001, 40, 284–310.
Otten, L. G.; Quax, W. J. Biomol. Eng. 2005, 22, 1–9.
Hibbert, E. G.; Baganz, F.; Hailes, H. C.; Ward, J. M.; Lye, G. J.; Woodley, J. M.;
Dalby, P. A. Biomol. Eng. 2005, 22, 11–19.
Reetz, M. T.; Prasad, S.; Carballeira, J. D.; Gumulya, Y.; Bocola, M. J. Am. Chem.
Soc. 2010, 132, 9144–9152.
20. Matosevic, S.; Micheletti, M.; Woodley, J. M.; Lye, G. J.; Baganz, F. Biotechnol.
Lett. 2008, 30, 995–1000.
21. Wang, Z.-F.; He, Y.; Huang, L.-J. Carbohydr. Res. 2007, 342, 2149–2151.
22. Brunton, N. P.; Gormley, T. R.; Murray, B. Food Chem. 2007, 104, 398–402.
23. Jansen, G.; Muskiet, F. A. J.; Schierbeek, H.; Berger, R.; van der Slick, W. Clin.
Chim. Acta 1986, 157, 277–294.
8
9
.
.
Svendsen, A. Enzyme Functionality – Design, Engineering and Screening; CRC
Press, 2003.
24. Sweeley, C. C.; Bentley, R.; Makita, M.; Wells, W. W. J. Am. Chem. Soc. 1963, 85,
2497–2507.
1
0. Goddard, J.-P.; Reymond, J.-L. Curr. Opin. Biotechnol. 2004, 15, 314–322.
25. Petersson, G. Carbohydr. Res. 1974, 33, 47–61.
1
1. Enzymatic coupling: The enzymatic coupling involves a long chain aldehyde
26. Protocol derivatization: After 24 h at 25 °C, 150 lL of the internal standard (IS)
acceptor (any sugar,
D
-ribose or
D
-glucose for example) and a keto acid donor
solution (glucose at 100 mM) was added just before the complete evaporation
of water using a SpeedVac concentrator. A large excess of hydroxylamine
(
such as lithium hydroxypyruvate). The protocol for this coupling can be
generalised to any substrate as follows: two solutions were prepared in a Tris–
HCl buffer (50 mM) at pH 7.0: the first, containing the cofactors at 18 mM for
hydrochloride was added (0.15 mL as a 2.5 M solution in pyridine). The
mixture was shaken and heated at 80 °C for 30 min. The sample was cooled to
room temperature before adding BSTFA, (+1% TMSCl) in excess (0.15 mL). The
reaction mixture was shaken at room temperature for 30 min. Finally, 0.7 mL
of EtOAc was added. After removal of the precipitated salts and by-products (if
necessary) by centrifugation, the sample was analyzed by GC.
2+
Mg (MgCl
00 mM each (aldehyde acceptor and keto acid donor). The 100
solution of cofactors was added to 100 L of lysate. The mixture was stirred for
0 min, and then 100 L of the solution of reagents was added, giving overall
2
Á6H
2
O) and 5 mM for TPP; the second containing the reagents at
1
l
L of the
l
1
l

A. Ranoux et al. / Tetrahedron Letters 53 (2012) 790–793
793
2
7. GC analysis: The analysis was carried out on a Shimadzu GC 2010 Plus
equipped with a Varian FactorFour VF-1ms column (length: 22.5 m; inner
Glucose and seduheptulose gave two signals, one for the cis and the other for
the trans isomer of the oxime derivative.
diameter: 0.15 mm; film thickness: 0.15
Shimadzu Auto-injector AOC 20i (carrier gas He). The detector temperature
was 330 °C, the injector temperature, 300 °C, the column flow 0.79 mL min
l
m) with
a
FID-detector, and
a
28. Standard solutions were prepared in Tris–HCl buffer (50 mM, pH 7) containing
both reagent and product (commercial compound if available or isolated
product) then derivatised and analysed by GC following the previously
described procedures.
À1
(
linear velocity 37 cm/s). A split ratio of 50 was used. The temperature gradient
began at 180 °C, was held for 12 min, then increased by 50 °C/min to 320, hold
29. Samland, A. K.; Rale, M.; Sprenger, G. A.; Fessner, W.-D. ChemBioChem 2011, 12,
1454–1474.
1
4
min. The total process required 16 min. The retention time of ribose was
.6 min, sedoheptulose 13.77 and 13.83 min, IS (glucose) 10.2 and 10.8 min.